EP0986928A1

EP0986928A1 - Dynamic channel assignment method in a cell communication network

Info

Publication number: EP0986928A1
Application number: EP98928413A
Authority: EP
Inventors: David Verrier; Patrick Plas
Original assignee: France Telecom SA
Current assignee: Orange SA
Priority date: 1997-06-04
Filing date: 1998-06-02
Publication date: 2000-03-22
Anticipated expiration: 2018-06-02
Also published as: DE69825892T2; WO1998056204A1; DE69825892D1; CN1130102C; CA2292738A1; EP0986928B1; JP2002502574A; CA2292738C; US6606499B1; JP4149524B2; FR2764464B1; FR2764464A1; CN1265255A; ES2226145T3

Abstract

The invention concerns a method whereby for each network base station each physical channel managed by the dynamic channel assignment (DCA) method is associated with a particular priority index computed on the basis of periodically measured radio parameters relative to communication intervening on this physical channel. For the assignment of a new communication channel, the free and accessible physical channels with the highest priority are given precedence. The priority indices can further be used for the dynamic transfer of current communications towards free and top priority channels, whereby the distribution of the pass band between cells as well the quality of communications is optimised.

Description

METHOD FOR DYNAMIC ALLOCATION OF CHANNELS IN A CELLULAR RADIO COMMUNICATION NETWORK

The present invention relates to dynamic channel allocation (DCA: "Dynamic Channel Allocation") for a cellular radiocommunication network such as a GSM type network.

A distinction is made between DCA techniques for adaptation to interference and DCA techniques for adaptation to traffic. Interference adaptation techniques take into account the quality of the radio signals received to dynamically assign the channels supposed to be the least noisy. Traffic adaptation techniques rely on knowing the channels used in each cell to ensure that two identical channels are not allocated in neighboring cells: they do not involve radio quality criteria, but require information exchanges between the different channel allocators in the network.

In the field of interference adaptation DCA algorithms, the algorithms hitherto exposed are suitable for channel allocation on the radio channel for packet mode communications (see M. FRULLONE et al: "Dynamic Channel Allocation for ATDMA ", Proc. Of the Race Summit, Lisbon, November 1995, pages 299-303). This is not suitable for networks which, like GSM, support circuit mode communications. As for DCA algorithms adapted to traffic, the studies carried out are essentially theoretical. Due to the fact that these mechanisms require significant signaling exchanges between the various entities allocating radio resources of the network, they are poorly suited to current cellular networks which do not facilitate these exchanges. They therefore present little interest for the moment.

Today, mobile network operators apply channel planning techniques to distribute the physical channels to be allocated over the different cells of the network. The term "planning" means that each cell is assigned a specific list of physical channels from which the channels are chosen when they are allocated. In most systems, channel planning is simplified into frequency planning.

The advantage of such a technique is that it is possible to ensure that no channel is allocated to two neighboring cells, which reduces the risk of interference between two allocated channels in neighboring cells. This ensures a certain radio quality for each channel allocated in the network.

However, frequency planning has the following disadvantages:

1) It is a tedious operation to perform. The more irregular the cell coverage topology (different cell sizes, non-symmetrical patterns, etc.), the more difficult it is to plan. The concept of multicellular network, with “umbrella” micro-cells and macro-cells, makes any attempt at frequency planning even more difficult, since it introduces several levels of coverage, each requiring frequency planning. This also has the effect of restricting the number of frequencies allocated to each cell (hence a traffic limitation).

2) Frequency planning does not facilitate changes in network topology. For example, each time base stations are added to or removed from the network, a new frequency planning is necessary on a large part of the network. This point is all the more important since operators are now often required to modify their cellular engineering (integration into the existing network of micro-cellular areas).

3) Frequency planning does not allow the system to allocate resources to mobiles flexibly. Due to the fixed (and limited) number of resources allocated per cell, the system cannot absorb local traffic surges due to lack of available resources. If frequency planning can remain an interesting solution in a macro-cellular environment (regular and simple cellular configuration, homogeneous traffic distribution), it is much less so for other types of environments (micro-cellular, “umbrella” cells). , traffic peaks, ...).

The strong current growth in mobile communications traffic due to the success of mobile phones is pushing operators to densify their network. These latter are then brought today to combine any type of cellular configuration (macro-cells, microcells, umbrella cells, omnidirectional antennas, directional antennas ...) and consequently demand a more flexible channel allocation mechanism than the plan frequencies.

WO96 / 31075 describes a dynamic channel allocation process for a cellular radio network, in which "statistical preferences", that is to say priorities, are assigned to different frequency channels in the same cell. The selection of a frequency channel to use is based on these "statistical preferences". The determination of "statistical preferences" is based on measurements of channel characteristics, carried out when these channels are not used.

The DCA mechanism of this document WO96 / 31075 makes no use of the radio measurements made during communication. The same applies to the mechanism described in US-A-5 507 008. According to this last document, the base station of a cell verifies, when establishing a communication, that the envisaged channel is not undergoing too much interference. If the channel / interferer ratio (CIR) is too low, the base station switches to the next channel in a general list common to all cells. GB-A-2 266 433 describes another DCA mechanism in which several frequency lists are kept per cell. The base station determines a loss of transmission of a signal returned by a mobile station to select a list of frequencies in which the channel is chosen according to a quality criterion. This quality criterion can in particular be based on the channel / inter erer ratio. Updating the frequency lists depends on the success of previous selection attempts on the channels concerned. Again, no use is made of radio measurements made during communication.

An object of the invention is to propose an efficient technique for allocating radio resources dedicated to communications which does not require frequency planning beforehand and which thus allows the operator to escape the above constraints.

It is further desired that the method allows the network to absorb traffic conditions under certain conditions.

The invention thus provides a method for dynamic allocation of channels in a cellular radiocommunication network, a network in which a set of physical channels is used to form logical channels dedicated to circuit mode communications between mobile stations and radio stations. geographically distributed base, each logical channel belonging to a physical channel, and in which, for each communication established between a base station and a mobile station on a logical channel, radio parameters representative of the conditions of said communication on said communication are measured periodically logical channel. According to the invention, the method comprises the following operations carried out for each base station: - associating with each physical channel of said set a respective priority index;

- keep a first list of physical channels that said base station is not using to communicate with a mobile station, and at least a second list of occupied physical channels each comprising at least one active logical channel dedicated to a communication in progress between said base station and a mobile station; updating the priority indexes associated with the physical channels of the second list on the basis of the radio parameters measured relative to the communications in progress on logical channels belonging to said physical channels; and

- when establishing a communication with a mobile station, select for said communication an accessible and non-active logical channel belonging to a physical channel whose priority index is maximum. Thanks to this DCA mechanism, it is no longer necessary to carry out frequency planning between the cells before the network is put into service in order to distribute between them the radio resources attributable to mobile communications. The fact that an operator no longer needs to carry out frequency planning then introduces great flexibility into network deployment. For example, the integration of microcellular layers into a macro-cellular network is greatly facilitated, since there is no longer any need to share the radio spectrum between the various cellular layers, and between the cells within the same layer.

The method makes it possible to obtain an automatic planning of the channels between the cells. It ensures that the system converges quickly and automatically to a stable configuration in which the radio resources are correctly distributed between the cells (no overlapping of channels between two neighboring cells). In addition, this DCA process is very reactive to the various modifications that may occur in the network (changes in topology, variations in traffic) because it can modify the distribution of channels between cells accordingly. The DCA method according to the invention can therefore be used in all types of configuration of a GSM or similar network. The invention has the advantage of not causing any modification of the current signaling protocols of a GSM type network. To apply the invention, it is sufficient to implement the DCA method at the BSC level.

(base station controllers). Although one of the advantages of a DCA mechanism is that it does not require a frequency plan (FCA), the DCA mechanism according to the invention is in no way incompatible with a frequency plan. A GSM operator can apply this DCA mechanism in a network that also uses a frequency plan. In fact, for each cell, it is possible to specify the extent of the radio spectrum from which the DCA algorithm will draw resources. It suffices for this to suitably choose the set of physical channels subjected to the process. The invention then allows all kinds of FCA / DCA combinations, which provides great flexibility in the deployment of the network.

In a preferred embodiment of the method according to the invention, for each base station, the channel priority indexes of the first and second lists are compared to trigger an automatic intracellular transfer of communication from a logical channel belonging to a physical channel occupied from the second list to an accessible logical channel belonging to a physical channel of the first list whose priority index is higher than that of said occupied physical channel of the second list.

This improves the speed of convergence of the DCA algorithm towards a stable configuration. But above all, this allows an optimization of the quality of communication on the channels used: if a cell tries to use a channel already used by a neighboring cell, it will quickly detect the presence of interference and will choose another channel by the intracellular handover procedure. Other particularities and advantages of the invention will appear in the description below of a nonlimiting exemplary embodiment, with reference to the appended drawings, in which:

- Figure 1 is a diagram of a cellular radio network implementing the present invention; Figure 2 shows data stored for each base station in the network of Figure 1; and Figures 3 to 6 are flowcharts of procedures applicable in a method according to one invention.

Figure 1 shows seven base stations (BTS) 10-

16 of a cellular radio network. The coverage area of each base station 10-16 is called cell C0-C6, and stylized by a hexagon in Figure 1.

In the remainder of this description, it will be assumed that the cellular network is a GSM type network, without this limiting the scope of the description. In a network of this type, each base station is linked to a functional unit called base station controller (BSC), each BSC being able to control one or more base stations. Thus, in the case represented in FIG. 1, the BSC 20 is associated with the base stations 10, 14, 15.

Each BSC is connected to a mobile service switching center (MSC) 21 serving in particular as an interface with the public switched telephone network. GSM systems use frequency division multiple access (FDMA) and time division (TDMA) mechanisms. Each physical radio communication channel is thus identified by a carrier frequency and a time slot index identifying the time position of the channel in the TDMA frame (8 slots per frame in the case of GSM).

The logical traffic channels (TCH) most often used to transmit speech or data use whole physical channels. Some types of dedicated logical channels, however, use only a fraction of a physical channel. This is the case, for example, for half-rate traffic channels which can be multiplexed in pairs on the same physical channel. Logical signaling channels called SDCCH ("Stand-alone Dedicated Control Channel") are used to convey call control, mobility management and radio resource management messages. The SDCCH channel is the first channel assigned to a mobile during call establishment. He is then released for make room for a TCH channel with its associated signaling channel (SACCH: "Slow Associated Control Channel") in the case of speech or data transfer services. However, for some services, the SDCCH channel can be maintained. This is particularly the case for short message transmission services. An M number of SDCCH channels can be multiplexed on the same physical channel (M = 4 or 8).

In the remainder of this presentation, reference will only be made to the full rate TCH channels and to the SDCHH / 4 channels which are most commonly used in the networks operated (M = 4). It will be observed that the procedures described can be easily extended to other types of logical channels such as half-rate TCH (M = 2), SDCCH / 8 (M = 8) ...

For each dedicated communication channel, the mobile stations and the base stations carry out measurements of radio parameters representing the conditions of this communication, in particular the power level received by the mobile station or the base station or the quality of the signal received by the mobile station or the base station. These measures are described in detail in GSM recommendation 05.08 (Draft pr ETS 300 578, 2nd edition, March 1995, European Telecommunications Standards Institute), to which reference may be made.

The measurements are carried out with a frequency linked to the SACCH multiframe (480 ms). For each communication direction, the RXLEV parameter is the average of the field levels of the samples received over the 480 ms period. Each value of RXLEV is coded from decibel to decibel on six bits, the value RXLEV = 0 corresponding to a power lower than - 110 dBm, and the value RXLEV = 63 corresponding to a power higher than 48 dBm. For each direction of communication, the quality parameter RXQUAL is deduced from the error rates of the bits received on the channel over the period of 480 ms estimated from the metric used in the Viterbi channel equalizer and / or in the Viterbi convolutional decoder. Each RXQUAL value is coded from 0 to 7 according to the intervals value where the observed bit error rate falls (respectively 0% -0.2% / 0.2% -0.4% / 0.4% -0.8% / 0.8% -1.6% / 1.6% -3.2% / 3.2% -6.4% / 6.4% -12.8% / 12.8% -100%). From RXQUAL = 4, we can say that the quality of the radio link becomes poor.

The measurements made by the mobile station on the downlink are included in a message called MEASUREMENT_REPORT in GSM terminology. For radio link control procedures, the base station transmits these measurements to its BSC in a message called MEASUREMENT_RESULT in which it also includes the measurements it has made on the uplink. These measurements are processed at the level of the BSC which performs the radio link monitoring functions. The present invention proposes to process these measurement samples received by the BSC as part of a dynamic channel allocation method. This process can be entirely implemented at the BSC level, so that it does not require any particular adjustment of the GSM protocols.

Each BSC 20 is associated with a memory 22 (FIG. 1) which contains lists of channels for each base station 10, 14, 15 that it controls. The structure of these lists L1, L2, L3 for each of the base stations is illustrated in FIG. 2.

The first list L1 contains physical channels which are unoccupied at the instant considered, that is to say that the base station is not using to communicate with a mobile station. To maintain the list L1, the memory 22 contains three arrays FI, Tl, PI of length at least equal to the number N of physical channels of the set of channels subjected to the DCA mechanism. If NI designates the number of unoccupied channels at the instant considered, each of these channels i (1 <i <NI) corresponds to the time slot Tl (i) (1 <Tl (i) <8) of the communication frequency Fl (i), and is associated with a priority index PI (i). These three tables are ordered in descending order of the priority indices PI (i). The second list L2 contains physical channels used as TCH traffic channels at the instant considered between the base station and a mobile station.

To maintain the list L2, the memory 22 contains three tables F2, T2, P2 of length at least equal to the number N of physical channels of the set of channels subjected to the DCA mechanism. If N2 designates the number of physical channels supporting a logical TCH channel active at the time considered, each of these channels j (1 <j <N2) corresponds to the time slot T2 (j) (1 <T2 (j) ≤ 8 ) of the communication frequency F2 (j), and is associated with a priority index P2 (j). These three tables are ordered in descending order of the priority indexes P2 (j).

The third list L3 contains physical channels supporting, at the instant considered, one or more logical channels SDCCH.

To maintain the list L3, the memory 22 contains five tables F3, T3, NB, LOC, P3 of length at least equal to the number N of physical channels of the set of channels subjected to the DCA mechanism. If N3 designates the number of physical channels supporting at least one SDCCH channel active at the instant considered, each of these channels k (1 <k <N3) corresponds to the time slot T3 (k) (1 <T3 (k) < 8) of the communication frequency F3 (k), and is associated with a priority index P3 (k). NB (k) represents the number of SDCCH logical channels supported by the k-th physical channel of the list L3 (1 <NB (k) <M), and LOC (k) locates the positions of these SDCCH channels on the physical channel . LOC (k) thus consists of four LOC bits (k, m) such that LOC (k, m) = 1 if an active SDCCH channel occupies the m th logical channel position (1 <m <M) of k- th physical channel of the list L3, and LOC (k, m) = 0 otherwise. These five tables F3, T3, NB, LOC, P3 are ordered in descending order of the priority indexes P3 (k). Each of the physical channels processed by the DCA method belongs to one of the three lists L1, L2, L3, and is therefore associated with a respective priority index Pl (i) or P2 (j) or P3 (k). These priority indexes are calculated and updated during radio communications intervening on the channels considered, that is to say while the channels in question are in the list L2 or in the list L3.

We will now describe, with reference to FIGS. 3 to 6, procedures implemented according to the invention for managing the lists L1, L2, L3 and dynamically allocating channels for communications.

In a GSM network, the method of allocating an SDCCH channel is at the choice of the manufacturer. Some prefer to assign an SDCCH channel to a completely free physical channel, anticipating that the latter will then be used by the TCH channel which will relay it. In this case, the method selects an accessible physical channel from the list L1 having a maximum priority index. By accessible physical channel is meant a physical channel which can be assigned to a transmit-receive unit (TRX) of the base station. A free physical channel FI (i), Tl (i) from the list L1 will be inaccessible if, for example, all the TRXs of the base station are already occupied on the time slot Tl (i).

In the case of a short message service, the above option has the disadvantage of reserving an entire physical channel only for an SDCCH channel for the duration of transmission of the message, that is to say that the bandwidth is not optimized.

A preferred option, an implementation mode of which is illustrated in FIG. 3, consists in choosing as a priority, for the allocation of a new logical channel SDCCH, a physical channel from the list L3 having at least one free SDCCH component. If such a channel is not available, the highest priority free and accessible physical channel will be selected from the L1 list.

With reference to FIG. 3, the search loop for an SDCCH channel among the physical channels of the list L3 begins at step 30 with the initialization k = 1. This index k is incremented by one unit at step 31 if k <N3 and NB (k) = M (successive comparisons 32 and 33). When comparison 33 reveals a channel k having at least one free SDCCH component (NB (k) <M), the BSC determines the position m of such a component (LOC (k, m) = 0) in a loop 34-36. The non-active logical channel corresponding to the position m of the physical channel F3 (k), T3 (k) is then selected in step 37 as the SDCCH channel to be allocated. The update 38 of the list L3 then consists simply in increasing by one the number NB (k) of SDCCH components occupied on the physical channel and in taking L0C (k, m) = 1. The BSC then passes to the procedure for managing the priority index of the physical channel supporting the logical channel, which will be described later.

When comparison 32 shows that no physical channel in the list L3 can be used to support the SDCCH channel to be allocated (k> N3), the BSC searches for the free and accessible physical channel of higher priority. The search loop is initialized by i = 1 in step 40. This index i is incremented by one unit in step 41 if i <NI and the time slot Tl (i) is inaccessible (successive tests 42 and 43 ). If comparison 42 shows that i> Ni, channel allocation fails due to the lack of available radio resources. In the normal case, test 43, which can consist simply in verifying that the number of physical channels already allocated and having the same time slot index as the channel tested is less than the number of TRXs of the base station, will reveal a channel free and accessible i. An arbitrary logical channel (for example that of rank 1) of this physical channel FI (i), Tl (i) is then selected in step 44 to constitute the channel SDCCH. The BSC then proceeds to update the lists L1 and L3. In step 45, it inserts into the list L3 the physical channel supporting the logical channel which has just been selected (with NB = 1 and LOC = 1000), increases by one the length N3 of the list L3 and reorders this list in order of decreasing P3 priority indexes. In step 46, the i-th channel of the list L1, which has just been selected, is deleted from this list whose length NI is reduced by one, and whose elements are reordered according to the priority indexes PI decreasing. The BSC then proceeds to the management of the priority index of the physical channel which has just been allocated.

With regard to the procedure for allocating a TCH traffic channel, GSM standards give operators certain freedoms. Thus, a TCH channel can be allocated at the first request for radio resources ("Very Early Assignment" method), or can succeed the establishment of an SDCCH channel. For reasons of optimization of radio resources, operators generally prefer to assign an SDCCH channel and then a TCH channel. In the latter case, it is possible to assign as the TCH channel for the communication either the physical channel which was already occupied by the SDCCH channel, or a free physical channel. The procedure for allocating a TCH channel illustrated in FIG. 4 makes it possible to take these different scenarios into account.

Its first step 50 consists in examining whether an SDCCH channel has been allocated prior to the establishment of the required TCH channel. There are two general cases where the TCH channel can be assigned without this succeeding the allocation of an SDCCH channel:

- the operator uses an allocation method of the "Very Early Assignment" type; or

- the establishment of the TCH is caused by an automatic transfer of communication between two channels

TCH ("handover" or HO).

If no SDCCH channel was allocated, the BSC executes a loop 40-43 similar to that previously described with reference to FIG. 3 to identify the accessible physical channel from the list L1 having the largest priority index. The assignment of the TCH channel fails if such a channel is not available. The i-th physical channel of the list L1, identified as being the free and accessible channel of highest priority, is then selected in step 51, then removed from the list L1 in step 52. Finally, at step 53, the selected physical channel is inserted into the list L2, which the BSC reorders in accordance with the priority indexes P2, and the number N2 of elements of this list is increased by one. The priority index of the channel that has just been selected is then managed according to the procedure described below.

If the initial test 50 shows that an SDCCH channel was allocated prior to the establishment of the TCH channel, it is examined during comparison 55 if the number NB (k) of active SDCCH components of the k th physical channel of the list L3 which supports the previously allocated SDCCH channel is equal to or different from 1. If it is different from 1, this number NB (k) is simply reduced by one unit in step 56 due to the closure of the SDCCH channel, then the BSC proceeds to steps 40-43 and 51-53 previously described to assign the free and accessible physical channel of higher priority.

If comparison 55 shows that the k-th physical channel of the list L3 only included the SDCCH channel allocated before (NB (k) = 1), the BSC identifies, during a loop 60-63 similar to loop 40 -43 the accessible channel of the list L1 having the largest priority index PI (i). In the absence of such a channel in the list Ll (i>

NI during a comparison 62), or if this maximum priority index PI (i) is not larger than the priority index

P3 (k) of the physical channel which supported the SDCCH channel

(P3 (k) Pl (i) during the comparison 64), then the BSC selects as channel TCH the k-th physical channel of the list L3 at step 65. In the next step 66, it removes from the list L3 the channel having just been selected, it decreases by one the number N3 of occupied channels of the list L3, and it reorders this list in the decreasing order of the priority indexes P3. The BSC finally proceeds to update the list L2 by inserting therein, in step 53, the channel which has just been selected.

If comparison 64 shows that P3 (k) <Pi (i), it is preferable to allocate the i-th channel of the list L1, the priority index of which is greater than that of the physical channel of the SDCCH. The BSC then removes the k-th channel from the list L3 in a step 67 identical to step 66, before proceeding to steps 51 to 53 to select the i-th channel from the list L1 and update the lists L1 and L2. FIG. 5 is a flow diagram of a procedure for managing the priority index P2 (j) of the j th physical channel of the list L2 on which a communication is in progress between a mobile station and the base station. This procedure is presented in the context of a TCH channel (list L2), it being noted that it is directly transposable in the case of an SDCCH channel (list L3).

As indicated previously, the updating of the priority indexes is based on the radio measurements carried out periodically on the uplink and downlink of the logical channel. A quality of the channel Q (t) is evaluated every n measurement samples, that is to say every n mul itrâmes SACCH. As shown in block 70 of FIG. 5, the quality Q (t) can be a function of the average values, denoted RXLEV _n , RXQUAL _n , of the field levels RXLEV and of the quality parameters RXQUAL contained in the last n samples of measure. The RXLEV and

RXQUAL averages can be those measured on the uplink, on the downlink, or in both directions of communication. The number n is for example equal to 4, which corresponds to an update of the priority indexes approximately every 2 seconds. An example of a function f (RXLEV _n , RXQUAL _n ) that can be used to calculate the qualities Q (t) is: f (RXLEV _n , 0) = +3 f (RXLEV _n , l) = +3 f (RXLEV _n , 2) = + 3-2. (RXLEV _n / 63) f (RXLEV _n , 3) = + l- (RXLEV _n / 63) f (RXLEV _n , 4) = -1-2. (RXLEV _n / 63) f (RXLEV _n , 5) = -5-2. (RXLEV _n / 63) f (RXLEV _n , 6) = -9-3. (RXLEV _n / 63) f (RXLEV _n , 7) = -12

In the next step 71, the BSC updates the priority index P2 (j) of the channel on the basis of the last calculated quality value, or more generally on the basis of the last q calculated quality values. The new priority index P2 (j) is for example equal to the average value of the last q calculated quality values. The function of update g indicated in block 71 of FIG. 5 is then: g [P2 (j), Q (t), ..., Q (tq)] = P2 (j) + [Q (t) -Q (tq)] / q, the priority indexes of all the physical channels being initialized to 0 at the start of the DCA process.

After updating 71 of the priority index P2 (j), the list L2 is reordered in step 72 in the order of decreasing priorities.

If the channel F2 (j), T2 (j), has been released before the acquisition of n new measurement samples (tests 75 and 76), the priority index P2 (j) cannot be updated. In this case, the BSC updates the lists L1 and L2 in steps 77 and 78 by moving the released channel from the list L2 to the list L1 (in the case of the release of a logical channel SDCCH whose physical channel comprises one or more other SDCCH channels, steps 77 and 78 are replaced by a simple decrementation of the number NB (k) and by the inversion of the corresponding bit of LOC (k)).

Figure 6 shows the flowchart of an optimization procedure for the DCA mechanism. This procedure aims to trigger intracellular handovers in order to optimize the use of the higher priority physical channels, which are in principle the most reliable. This procedure controls two main actions: - when a logical channel is released, if the physical channel it occupied has a higher priority than that of a physical channel in use by a logical channel of the same type, then the BSC triggers an intracellular handover for this last logical channel towards the released channel whose priority is higher,

- when the quality of a physical channel seriously degrades, the procedure initiates for each of the logical channels that it supports an intracellular handover towards another physical channel as soon as the priority of the physical channel in use becomes lower than the priority another physical channel that can support the logical channel.

The operations performed in response to the initiation of an intracellular handover are described in detail in the aforementioned GSM recommendation 05.08, and will not be described in detail here.

The optimization procedure can be used for traffic channels as well as for SDCCH channels. However, since the duration of use of the SDCCH channels is generally short and since several of these channels can be multiplexed on the same physical channel, it seems preferable to apply the optimization procedure only to the traffic channels. This avoids a large number of intracellular handovers of SDCCH channels.

A clock is used so that the optimization procedure is only executed at regular time intervals (period T). At the end of this period, that is to say at the end of the timing step 79, the tables PI and P2 containing the ordered priority indexes of the physical channels of the lists L1 and L2 are recorded in a working memory (step 80), so that the optimization procedure uses fixed lists so as not to be disturbed by the updates that may occur during its execution in accordance with the procedure described with reference to FIG. 5. The period T can be limited to a few tens of milliseconds (for example T = 20 ms) to avoid this fixing of the priority indexes in step 80 leading to work on obsolete data. After registering the priority indices PI, P2,

The pointing index i in the list L1 is initialized to zero in step 81. In the next step 82, this index i is incremented by one, and the pointing index j in the list L2 is initialized at length N2 of this list. The priority indexes of the i th channel of the list L1 and of the jth channel of the list L2 are compared in step 83. If Pl (i)> P2 (j) + Δ, the BSC will trigger an intracellular handover from channel F2 (j), T2 (j) to channel FI (i), Tl (i) in step 85, provided that test 84 shows that the time slot Tl (i) is accessible to one TRX taking into account the possible release of the T2 tranche (j). After triggering the handover, the BSC updates the L1 and L2 lists by swapping the channels that were the subject of the handover, and by sorting the lists again in order of decreasing priority indexes, in step 86.

The margin Δ can be equal to 0. It can also be greater than 0 if one wishes to avoid triggering handovers which would only provide a small gain in quality.

If the comparison 83 shows that PI (i) <P2 (j) + Δ, the pointing index j in the second list L2 is compared to the length N2 of this list during step 88. If j = N2, this is because the list L1 does not contain any physical channel whose priority index would be sufficient for a handover to be interesting from the physical channel used having the lowest priority index. In this case, the optimization procedure ends with a return to the timeout step 79. If the comparison 88 shows that j <N2, the BSC returns to step 82 in order to examine whether the next channel of the list Ll would be a good candidate for an intracellular handover.

If test 84 shows that the i-th free physical channel of the list L1 is not accessible to receive a communication in progress on the j-th channel of the list L2, the pointing index j in the list L2 is compared to 1 in step 89. If j> 1, this index is decremented by one unit in step 90 before the BSC performs a new comparison 83. If comparison 89 shows that j = 1, the index The pointing i in the list L1 is compared with the length Ni of this list in step 91. If i = NI, it is because the scanning of the list L1 has revealed no accessible channel of sufficient priority, so that the optimization procedure is terminated by a return to the timing step 79. If the comparison 91 shows that i <NI, the BSC returns to step 82 in order to test the next channel of the list L1.

Note that the procedure of FIG. 6 authorizes only one intracellular handover for each period T, which is generally sufficient taking into account the relatively low value of T. However, it would be possible to trigger several handovers during this period between channel pairs meeting the comparison criterion 83.

The above optimization procedure is relatively simple to implement, especially since, in the vast majority of cases, the BSC will follow in the flowchart either path 80-83, 88, 79, or the path

80-86, 79.

It will be noted that there is a certain latitude in the definition of the concept of physical channel with which a priority index is associated in the method according to the invention. In the above description, reference has been made to the notion of the most natural physical channel for the TDMA / FDMA structure of GSM systems, namely that a physical channel corresponds to a carrier frequency FI (i), F2 (j) , F3 (k) and a time slot number Tl (i), T2 (j), T3 (k). In the case, for example, of a half-speed GSM system, it is possible to consider, for the management of the priority indexes, physical channels of half capacity. Multiple other conventions would be possible.

Claims

1. Method for dynamic channel allocation in a cellular radiocommunication network, a network in which a set of physical channels is used to form logical channels dedicated to circuit-mode communications between mobile stations (MS) and base stations (10-16) geographically distributed, each logical channel belonging to a physical channel, and in which, for each communication established between a base station and a mobile station on a logical channel, it is measured periodically radio parameters representative of said communication on the logical channel, the method being characterized by the following operations carried out for each base station: - associating with each physical channel of said set a respective priority index (PI (i), P2 (j), P3 (k) );

- maintain a first list (L1) of physical channels that said base station is not using to communicate with a mobile station, and at least a second list (L2, L3) of occupied physical channels each comprising at at least one active logical channel dedicated to a communication in progress between said base station and a mobile station;

- updating the priority indexes associated with the physical channels of the second list (L2, L3) on the basis of the radio parameters measured in relation to the communications in progress on logical channels belonging to said physical channels; and

- when establishing a communication with a mobile station, select for said communication an accessible and non-active logical channel belonging to a physical channel whose priority index is maximum.

2. Method according to claim 1, in which, for each base station, the priority indexes (PI (i), P2 (j)) of channels of the first and second lists (L1, L2) are compared to initiate a transfer automatic intracellular communication from a logical channel belonging to an occupied physical channel of the second list to an accessible logical channel belonging to a physical channel of the first list, the priority index of which is greater than that of said occupied physical channel of the second list.

3. Method according to claim 1 or 2, wherein the physical channels of each list are arranged in 1 order of decreasing priority indexes.